NETWORK OF REMOTE SENSORS FOR MAGNETIC DETECTION

Transcription

1 NETWORK OF REMOTE SENSORS FOR MAGNETIC DETECTION A. Sheiner 1, N. Salomonsi 1, B. Ginzburg 1, A. Shalim 1, L. Frumis, B. Z. Kaplan 1 R&D Integrated Systems Section, Propulsion Division, Soreq NRC, Yavne 81800, Israel Department of Electrical and Computer Engineering, Ben-Gurion University of the Negev P.O. Box 65, Beer-Sheva 84105, Israel Abstract A netor of remote magnetic sensors is used for detection of ferromagnetic objects. The magnetic sensors are arranged in gradiometric pairs in order to cancel bacground noise. For an effective noise cancellation the sensors' readings are synchronized by a smart synchronization technique. Data from the sensors are transmitted via a ireless lin, managed by an intelligent access protocol. The data are collected by a ireless acquisition unit hich transfers it serially to a monitoring station for further processing. The monitoring station consists of a PC that runs an advanced detection algorithm hich extracts a signal of magnetic dipole out of a bacground noise. An additional feature is the ability to distinguish beteen targets and false alarms. ith a 50 Hz / 60 Hz digital notch filter, RF communication controller, and a battery. The data from the sensors are collected by a ireless acquisition unit, hich transfers it serially to a monitoring station for further processing. The monitoring station consists of a PC that runs an advanced algorithm for detection of the magnetic anomaly. The system architecture is depicted in figure 1. Keyords: Magnetic Anomaly Detection (MAD), Sensor netor. 1. Introduction Magnetic Anomaly Detection (MAD) [1] is a passive technique for detection of ferromagnetic objects. The magnetic field produced by the object is modeled as a field of a dipole hich generates an anomaly in the ambient Earth magnetic field. We have developed a netor of remote magnetic sensors in order to detect the anomaly in Earth magnetic field. The sensors are arranged in gradiometric [] pairs in order to cancel common components of far magnetic noise sources such as geomagnetic noise []. Each remote sensor consists of a single-axis fluxgate magnetometer, an interface module Figure 1 Sensors netor for remote detection of ferromagnetic objects. Magnetic Anomaly Detection Farther aay from the object, the magnetic field produced by the object may be considered as a magnetic dipole field: r r r r ( m r) r r r µ 0 r m B( m, ) = 5 4π r r (1)

2 here B r is the magnetic field that is produced by an object ith a magnetic moment m r at displacement r. Here, B r can be treated as anomaly in the ambient Earth magnetic field, B r E, as is given by: r r r B = B+ () m B E. The Remote Magnetic Sensing Unit The Remote Magnetic Sensing Unit (RMSU) consists of a single-axis fluxgate magnetometer, an interface to the magnetometer ith a 50 Hz/ 60 Hz digital notch filter, RF communication controller and an energy source. The architecture of the RMSU is depicted in figure. direction of Earth magnetic field, the readings of a single-axis magnetometer resemble the readings of a total field magnetometer []. The chosen single-axis magnetometer has to analog outputs. A direct output hich the voltage is proportional to the sensed magnetic field, and a filtered output hich is obtained by processing the direct output by a Band Pass Filter (BPF), in order to eliminate the DC component. The RMSU is equipped ith a lo poer narro band RF transceiver. Communication range of fe hundred meters may be achieved using a λ/4 antenna. In order to extend communication range, a higher gain antenna is available. The output of the magnetometer is sampled at a rate of 500 samples per seconds. The samples are processed through a digital notch filter in order to suppress 50Hz / 60Hz interferences. The filter decimates the samples don to a rate of 10 samples per second, hich is sufficient for object detection as ell as for gradiometric noise cancellation. The decimated samples are stored in a buffer until a data request command is received, then the buffer data are transmitted over the RF communication lin. Careful mechanical and electrical design as made in order to eliminate magnetic interferences caused by the modules of the RMSU. 4. The Wireless Acquisition Unit Figure The inside structure of the RMSU Due to poer saving considerations e preferred to use a single-axis magnetometer rather than a three-axis magnetometer, although the latter one provides more information about the target. The singleaxis magnetometer has been installed in the direction of Earth magnetic field in order to reduce magnetic interferences caused by orientation instability. When installed in the The Wireless Acquisition Unit (WAU) coordinates the data flo from the RMSUs using an advanced access protocol that as designed to meet the particular needs of the system. For effective bacground noise cancellation the readings from the RMSUs should be synchronized. The synchronization is achieved through the advanced access protocol of the WAU. The WAU scans the RMSUs successively by initiating a data request command via the RF communication lin. Only the RMSU ith the addressed ID responds by transmitting its buffer data over the RF communication lin. The communication beteen the WAU and the RMSU is organized in pacets hich include a preamble, a synchronization

3 symbol, header, filtered readings from the magnetometer, and a checsum byte. The WAU is configured to resend immediately a second data request command in the case of a pacet loss. If there is no valid data reception after resending data request command, then in the next scan a double amount of data ill be sent from the RMSU. A synchronization command is generated periodically by the WAU for all the RMSUs in order to synchronize all RMSUs clocs. The data from all the RMSUs are collected by the WAU and transferred serially for further processing by the monitoring station. 5. The Monitoring Station The monitoring station consists of a PC running a softare application, hich performs several routines: (a) It receives serially all RMSUs data that ere collected by the WAU (b) The data from all the RMSUs are arranged in gradiometric pairs ith time accuracy better than 0.1 sec. (c) An advanced algorithm is applied to the synchronozied data in order to detect a magnetic anomaly. The algorithm relies on signal decomposition into three basis functions [4]. Consider the case of a static scalar magnetometer and a magnetic object hich moves along a straight line. The sampled signal may tae various forms, depending on object magnetic moment orientation and trac. The diversity of possible shapes of the measured signal prevents the use of detection techniques based on pre-non signal shape. Therefore, an advanced detection method is needed such as the Orthonormal Basis Function (OBF) detection method [5]. Assuming that the mobile object moves along a straight line, e define as the ratio beteen the position of the object along its trac, d and R 0. R 0 is the closest point from the target to the sensor trac, entitled as the Closest Proximity Approach (CPA). Under this assumption, e can present the object magnetic signal as a function of the single argument. It can be shon that the magnetic target signal is composed from a set of three OBFs, f ( ) and coefficients c, for =1,, : ( ) = B c f ( ) () = 1 Implementing the Gram-Schmidt procedure yields the folloing OBFs: f1( ) =.5 5π ) f f ( ) = ( ) = π 18 π ) ).5 (4) Folloing optimal detection theory [6], a matched filter is applied by taing a indo of L samples from the measured signal and correlating the indo ith each of the OBFs. The OBFs should also be arranged in vectors of L elements each. The outputs of the three correlations are then squared and summed in order to form a decision index. Detection occurs hen the index value exceeds a predetermined threshold. The process is depicted in figure. For a distant object the signal is expected to be ider than for a close object. Hence, the indo length L should be determined according to R 0 value. Therefore, a guess must be made for the value of R 0, or a multi-channel approach for various values of R0 could be adopted. 6. Further investigation Preliminary tests indicate good detection capabilities. Further efforts could be made in order to extend system performance. The detection algorithm is based on the matched filter theory, hich is optimal for detection of a non signal contaminated by hite noise. Hoever, measured noise features poer spectral density (PSD) of 1/f α [7], here usually 0<α<. The 1/f α noise is due to the nature of geomagnetic noise and the intrinsic characteristic of the sensor. A

4 hitening filter may be used to improve detection by OBFs in presence of 1/f α noise. In many cases a detection of a magnetic object should be folloed by characterization of the object [8]: localization and magnetic moment estimation. Methods such as the Genetic Algorithm (GA), hich ere tested by simulation for magnetic target characterization [9], may be applied for this purpose. 7. Conclusions We have developed and tested a netor of remote sensors for magnetic anomaly detection. For effective noise cancellation, the readings of the sensors must be synchronized. Hence a smart ireless handshae as developed to ensure synchronization. Preliminary tests ere carried out, shoing promising results. A system prototype is being evaluated in outdoor conditions. The system is designed for several applications such as: multi-sensor portable system for precise magnetic measurements, advanced intrusion detection system, car traffic monitoring [10], etc. Additional efforts ill be made in order to extend system capabilities. 8. Acnoledgements We ould lie to than Mrs. Avivit Noyman, Mr. Yossi Ele, and Mr. Yohai Nitsan, for their important role in system engineering. Figure Digital Implementation of a magnetic target detection filter using OBFs 4

5 9. References [1] Wynn W. M., "Detection, Localization, and Characterization of Static Magnetic Dipole Sources", Detection and Identification of Visually Obscured Targets, (1999), [] Ginzburg B., Frumis L., and Kaplan B. Z., "An efficient method for processing scalar magnetic gradiometer signals", Sensors and Actuators A, Vol. 114, (004), [] Ash A. D., "Noise and Noise Reduction Techniques for Airborne Magnetic Measurements at Sea", International Conference on Marine Electromagnetics, MARELEC, (1997). [9] Sheiner A., Salomonsi N., Ginzburg B., Frumis L., and Kaplan B.Z., "Aeromagnetic Search using Genetic Algorithm", Progress in Electromagnetic Research Symposium, PIERS, (005), [10] Casalegno J. W., "All-Weather Vehicle Classification Using Magnetometer arrays", Unattended Ground Sensor Technologies and Applications, Proceedings of SPIE, Vol. 474, (00), [4] Ginzburg B., Frumis L., and Kaplan B. Z., "Processing of magnetic scalar gradiometer signals using orthonormalized functions", Sensors and Actuators A, Vol. 10, (00), [5] Frumis L., Ginzburg B., Salomonsi N., and Kaplan B. Z., "Optimization of scalar magnetic gradiometer signal processing", Sensors and Actuators A, Vol. 11, (005), [6] Hayin S., "Adaptive Filter Theory", rd edition, Prentice Hall, (1996), [7] Davis J. P., Gelata W., Allen O.,Bobb L. C., "Preliminary Tests of the Multi- Mode Magnetic Detection System", Proceedings of the International Conference on Marine Electromagnetics, MARELEC, (004). [8] Ginzburg B., Sheiner A., Frumis L., and Kaplan B. Z., Salomonsi N., "Investigation of advanced data processing technique in magnetic anomaly detection systems", International Conference on Sensing Technology, ICST, (005),